DAEJEON — The Korea Advanced Institute of Science and Technology (KAIST) is strengthening its teaching in literature, history and philosophy to train scientists and engineers who can think for themselves in the age of artificial intelligence. The aim is to ensure that engineers leading technology development carry a sense of social responsibility and ethical judgment at a time when AI's role has expanded beyond delivering knowledge to waging war.
"Engineering technology is deeply connected to life, safety and the environment," KAIST President Bae Choong-sik said in an interview with The Seoul Economic Daily at the university's main campus in Yuseong District, Daejeon, on the 15th. "We absolutely need a vision of the human being as an independent agent, grounded in philosophy." He added: "If people doing engineering are not armed philosophically, they can end up fixated on efficiency alone. We are recruiting more professors in the basic humanities and social sciences, and we are asking the relevant departments and colleges to strengthen teaching in literature, history and philosophy."
Bae graduated from Seoul National University's Department of Aerospace Engineering and earned a doctorate in mechanical engineering from Imperial College London. He joined KAIST in 1998, served as dean of the College of Engineering and other posts, and took office as the institute's 18th president in July this year. Having set out a vision of a "globally innovative leading university with strong fundamentals," he is reorganizing education, research and administration around AI while placing weight on building students' ability to understand principles and make judgments on their own.
In the interview, Bae voiced strong concern about the current educational environment, in which AI takes over the learning process. If AI does the reading, thinking and writing, he said, assignments may come back polished while students' own reasoning goes undeveloped — making a shift away from existing teaching methods necessary.
His proposed answer is "AI-free" education. Students prepare in advance, then in the classroom are trained to think, write, speak and present on their own without help from AI or the internet. Whether AI may be used depends on the purpose of the class. Even in the natural sciences and engineering, its use is restricted in activities where students must understand basic principles and explain them in their own logic. By contrast, in activities applying knowledge from their major to real problems, AI is used actively, but students must verify the errors and validity of the results themselves. The idea is to build, at the same time, the ability to distinguish what can be left to AI from what must be judged for oneself.
Philosophy education is the foundation of this overhaul. "If engineers cannot judge the purpose and social impact of technology, the pursuit of performance and efficiency can produce side effects for life, safety and the environment," Bae said. "We are designing an approach that links education that builds reasoning and basic principles without AI, education that uses AI to solve problems, and education that designs and develops AI itself."
In research, KAIST is putting its weight behind physical AI, drawing on South Korea's manufacturing base. Bae assessed that the country holds a "full stack" advantage, with a range of talent and an industrial structure spanning the basics through applications. "If we secure the foundational capabilities we need through sovereign AI while applying AI to manufacturing, we can reach a level in physical AI that the world takes as its model," he said.
Doing so requires connecting AI's judgments to the actual movement of machines and to production processes. "Beyond algorithms and semiconductors, you need expertise in machine control, materials and processes," Bae said. "If we combine KAIST's research capabilities across the natural sciences, electronics and computing, and mechanical, chemical, materials and bio engineering, we can put forward leading physical AI models and products." Under the plan, the existing College of AI serves as the hub while the AI research institute links faculty across fields, with industrial application pursued through industry-academic research with companies such as Nvidia.
The institute is also setting up an education track that lets students take startup-related courses on an elective basis, from before enrollment through graduate school, to systematize teaching so that research results lead to new ventures. The goal is to reorganize courses that are now fragmented and to map out learning paths by year, helping students take the leap without hesitation once they have technology worth commercializing.
Early in their studies, students will also be encouraged to consider entrepreneurship as one career path by hearing from successful alumni. "To that end, I have recently been meeting alumni whose names people would recognize and asking them to take part in teaching for younger students," Bae said. "After that, we guide students to build the grounding a business requires — economics, management, accounting, taxes and law — and we are also considering a certification scheme so they can list completion of the track on their resumes."
Expanding infrastructure is critical to supporting such education and research. "Computing resources are short, so along with securing GPUs we are weighing 'networking' to raise the utilization of existing resources," Bae said. Networking means linking up and using computing facilities at outside companies, or letting one laboratory use resources another laboratory is not using. The idea is to offset limits on the funding and electricity needed to build an in-house data center by sharing resources; concrete cooperation with companies is still at the planning stage, he said.
Bae also called for an environment outside the university that allows people to try again after failure. "Because a startup is a challenge with no guarantee of success, we need systems and a culture in which young people can get back up even after failing several times," he said. "I hope they will be allowed to pursue research and education that is free, bold and future-oriented."